🐍📦 pyppbox#

pyppbox is an all-in-one Python toolbox which can be used for detecting, tracking, and re-identifying people with only a few lines of code. It is originally made for PoseTReID framework which can effectively track specific/certain people across multiple cameras or video scenes in distributed contexts of people interaction spaces such as malls or amusement parks, etc.

  • pyppbox = Python + People + Toolbox

  • Design for both short and long term people detecting, tracking, and re-identifying.

  • Integrate GUI for easy configuration and demo.

  • Support dictionary and YAML/JSON configuration.

  • Support [GTA_V_Dataset] -> Real-time online and offline evaluation.

  • Check our papers: 1 [IEEE] | 2 [IEEE] [arxiv]

pyppbox 4.0.0#

Version 4.0.0 adds RF-DETR detection, ArcFace/AdaFace recognition, and laprust assignment solving. It improves pipeline cleanup, configuration validation, and result exports, while preserving existing V3 configuration files.

Install or upgrade to the general release, including from a release candidate:

python -m pip install --upgrade "pyppbox==4.0.0"

Core compatibility covers Python 3.9–3.13. RF-DETR, ArcFace, and AdaFace require Python 3.10+ and optional dependencies. Install the model/data wheels for your selected modules using the setup guide. See the V4 upgrade guide for dependency changes, UTF-8 text files, and lifecycle APIs, or the release notes for the full changes.


Table of contents#


Indices and tables#